In the U.S. EPA’s regulatory model (MOVES), fuel use rates are functions of on-road operating conditions (speed and acceleration conditions). The model uses operating mode (OpMode) distributions, defined by vehicle specific power (VSP) or scaled tractive power (STP), speed and acceleration, to represent the time fraction of engine load. MOVES converts driving cycles into OpMode distributions for project-level analysis (using MOVES default or interpolated driving cycles, customized cycles from user inputs, or OpMode distribution inputs). However, the relationships between average speed and fuel use rates are non-linear. Hence, it is critical to understand how various input variables (e.g., source type, model year, ambient temperature, and relative humidity) impact the sensitivity of these rates against average speed. The release of MOVES5 includes updates to algorithms since MOVES2014 (including VSP and STP parameters) and introduces base rates that are expected to alter these sensitivity profiles compared to the previous version, MOVES4.0.1. The research team will perform a comprehensive sensitivity analysis of the speed-fuel use relationships using the high-performance computing capabilities of MOVES-Matrix 5.0 (which generates identical results to the MOVES model but runs 200 times faster). The team will investigate the impacts of input variables on these sensitivities and explicitly compare the results against MOVES4.0.1 to document how model updates may impact planning outputs. Fuel use analysis on freeway corridors also depends on the interaction between traffic volume and operating speeds. Because project-level analyses often aggregate data over time (e.g., one hour) and space (e.g., lane-by-lane to corridor), there is a risk of losing modal model fidelity under variable congestion conditions and variable fleet compositions. The team will perform an uncertainty analysis associated with the temporal and spatial aggregation of on-road operations data (flows and speeds) within the MOVES 5.0 framework, comparing these findings to prior studies to assess how data resolution impacts the accuracy of estimation within the simulator.